Risk perception, but also political orientation, modulate behavioral response to COVID-19: A randomized survey experiment

Abstract

Prior work has shown that accurately perceiving the risk for COVID-19 is associated with higher adherence to protective health behaviors, like face mask use, and more acceptance of governmental restrictive measures such as partial or complete banning of indoor activities and social gatherings. In this study we explored these associations at the beginning of the second wave of COVID-19 in Argentina through a national representative probabilistic survey that evaluated personal and contextual risk perception, self-reported compliance with protective health behaviors, attitude to governmental restrictive measures, and political orientation and psychological distress as potential modulators. Also, going beyond measures of association, here we sought to test whether messages highlighting potential risks increased acceptance of restrictive measures. Three types of messages were randomized to the participants. Two messages conveyed risk-related content (either through emotional arousal or cognitive appraisal) and the third a prosocial, altruistic content. Between March 29th and 30th, 2021, 2,894 participants were recruited (57.57% female). 74.64% of those surveyed evaluated the current health situation as "quite serious" or "very serious" and 62.03% estimated that the situation will be "worse" or "much worse" in the following 3 months. The perception of personal risk and the level of adherence to protective behaviors gradually increased with age. Through a regression model, age, perceived personal risk, and contextual risk appraisal were the variables most significantly associated with protective behaviors. In the case of the acceptance of restrictive measures, political orientation was the most associated variable. We then found messages aimed at increasing risk perception (both emotionally or cognitively focused) had a significantly greater effect on increasing the acceptance of restrictive measures than the prosocial message, mainly for government supporters but also for non-supporters. However, the level of response was also modulated by the political orientation of the participants. We propose a mechanism of "ideological anchoring" to explain that participants were responsive to risk modulation, but within the limits established by their pre-existent political views. We conclude that messages highlighting risk can help reinforce the acceptance of restrictive measures even in the presence of polarized views, but must be calibrated by age and political orientation.

Keywords: COVID-19; anchoring; health psychology; lockdown; political orientation; protective health behaviors; psychological distress; risk perception.

FIGURE 1

Pairwise scatter plots between variables. Locally weighted linear regression lines are added for illustrative purposes of how the data is distributed, not for statistical inference. The univariate distributions are displayed on the diagonal.

FIGURE 2

Compliance with protective health behaviors. (A) Percentage of participants compliant with protective health behaviors according to age group. (B) Percentage of participants compliant with protective health behaviors according to political orientation. Error bars represent bootstrapped 95% CI.

FIGURE 3

Support for government restrictive measures. (A) Percentage of participants supporting the different levels of restriction according to age groups. (B) Percentage of participants supporting the different levels of restriction according to political orientation. Error bars represent bootstrapped 95% CI.

FIGURE 4

Effect of messages on restrictions acceptance. (A) Participants’ responses to the three types of messages in the total sample. (B) Participants’ responses to the three types of messages among the government supporters. (C) Participants’ responses to the three types of messages among the opposition supporters. Error bars represent bootstrapped 95% CI (COG, cognitive risk message; EMO, emotional risk message; SOC, pro-social message; Complete response: “Support all necessary restrictions”; Partial response: “Support some additional restrictions”; Negative response: “Support only the current restrictions” or “No restrictions at all”).

FIGURE 5

Predicted vs. observed outcomes in ordinal regression. Red dots correspond to matched cases (Pro-G, government supporters; Opp, opposition supporters).